6.9
Optimal Management Zone Delineation
Investigators
Carl
R. Dillon, Agricultural Economics, cdillon@ca.uky.edu
Scott A. Shearer, Biosystems and Agricultural Engineering, shearer@bae.uky.edu
Tom Mueller, Agronomy, mueller@uky.edu
Cooperators
Mike
Ellis, Kentucky Producer and Precision Agriculture User, wdemike@iglou.com
Introduction
One
of the most basic and perplexing questions involved in variable rate technology
within precision agriculture is the delineation of the profit maximizing
management zones which are treated separately with respect to input application.
While more accurate information regarding the optimal level of input
(e.g., fertilizer) is desirable on a fine scale, the fixed costs per zone or
grid (e.g., soil sampling) associated with this greater accuracy will not
justify the additional costs at some point.
The decision of how to delineate optimal management zones or grid sizes
represents a great opportunity for profit while at the same time presenting a
daunting and complex problem that confuses researchers, extension specialists,
industry leaders and producers alike. Consequently,
producers are left facing the difficult decision of how to delineate management
zones without suitable guidance. Alternatively,
some producers desire to establish uniform grid size (as opposed to variably
sized and shaped management zones) and face the question of the best grid size
to use. Although some informal
standards regarding the 2.5 acre grid size is often used, decision tools are
needed to provide a more robust management zone delineation procedure and
uniform grid size level projections for farmers using variable rate application
of inputs. The focus of this
proposed research is upon this very fundamental issue of precision agriculture
that is so critical to adequate economic implementation of variable rate
application regardless of the input (seed, fertilizer, pesticide).
The most appropriate method of responding to
this research question is through a multidisciplinary framework.
Specifically, a model embodying the decision-making framework of the
producer will allow for proper analysis of these questions.
A model that allows for the objective of maximization of profits subject
to the constraints a producer faces reflects the production environment faced by
the farm manager. Ultimately, a
farmer’s decision is derived from the underlying economic consequences of the
potential courses of action being considered. In turn, the economic consequences
are determined by the underlying production responses.
Therefore, agricultural economic results drive a producer’s decisions
while the physical agricultural relationships (e.g., agronomic, engineering)
provide the foundation for the economic results. Additionally, this basic framework can be expanded to
incorporate consideration of risk to more completely ascertain economic
sustainability.
While this work will be primarily used to
assist corn, soybean and wheat crop producers, the techniques developed here
will be suitable for a broader audience. Because
this work provides a missing key element for properly assessing alternatives
regarding variable rate input application, this research is especially relevant
to all row crop producers who use variable rate technology.
Objectives
The fundamental purpose of this research
project is to assist crop producers by providing procedures and information that
will assist them in making sustainable decisions concerning how to optimally
delineate management zones. The
procedures developed here can be generalized to several production inputs (e.g.,
N, P and K). The specific
objectives include:
1.
Identify and develop various procedures that permit the determination of
the specific optimal regions to be included within a management zone,
2.
Assess the economic performance of these procedures to compare their
relative success to the profit maximizing management zone as well as communicate
these results to the public and provide decision aids for the public’s use,
and
3.
Perform sensitivity analysis to ascertain how optimal management zones
change/respond to fluctuations in the economic decision-making environment
including consideration of risk.
Background
Background
information in the form of a review of literature can serve to establish a basic
framework for the proposed study. Included in this will be a general discussion
of economic studies pertaining to precision agriculture. Studies that examine
variable rate technology, grid sampling and management zones are then discussed
to complete the background information.
The economic feasibility of precision
agriculture is a common underlying question of producers considering its
adoption. While the
literature regarding the economic issues in the area of precision agriculture is
rich with numerous studies, they are broad based and display a substantial
number of philosophical discussions rather than quantitative evaluation as is
common with new technologies. General
philosophical discussions have ranged from historically descriptive (e.g.-
Lowenberg-DeBoer; Sonka and Coaldrake) to examining the research opportunities
and challenges of the future (e.g.-Weiss).
Lowenberg-DeBoer and Swinton conducted a review of the economics of
precision agriculture, finding that economic feasibility is dependent upon
several factors including many components of the underlying economic, agronomic
and engineering environment. Precision
agriculture has been shown to be profitable, not profitable or inconclusive with
mixed results, depending on the crop, inputs and conditions.
In addition to these diverse precision
agriculture economic studies, three specific areas are worthy of attention for
this research: variable rate technology, grid sampling and management zones.
Variable rate technology research has included analysis of such
components as nitrogen management (e.g., Thrikawala et al.; Babcock and Pautsch),
lime application (e.g., Bongiovanni and Lowenberg-DeBoer) and spatial break-even
variability assessment (English, Roberts and Mahajanashetti).
Studies rely predominantly upon the use of an assumed level of grid
sampling, avoiding the issue of optimal grid size or management zone
determination with few exceptions (Thrikawala, Weersink and Kachanoski). While some economic research investigates grid sampling
issues (e.g., Lenz; Rehm et al.), there is a void in the literature for sound
economic models to address optimal grid size which is exceeded only by the
apparent lack of economic analysis of the determination of optimal management
zone delineation.
The research proposed in this project lies
at the very heart of precision agriculture. It will assist in the establishment
of a fundamental framework that will permit analysis of a very germane and basic
question currently plaguing the successful implementation of variable rate
technology. Specifically, how does
a producer identify the optimal management zone?
The innovative model formulation proposed within this research project
permits the appropriate economic analysis to be conducted for comparison to the
less data intensive and more farmer friendly management zone delineation
procedures presently being conceived and tested by others (such as delineation
by soil properties or electrical conductivity). Therefore, this research also provides complementary economic
comparison among alternative procedures for management zone delineation.
Additionally, the techniques include the economic assessment of another very
important question: What is the optimal uniform grid size?
Thus, while some producers will use the variably sized and shaped
management zones and others will use uniform sized and shaped grids, optimal
determination of both can be handled with the model proposed.
Furthermore, this study will provide insights into the establishment of
practical and simple decision rules.
Procedures
The
objectives of this study will be accomplished in three steps.
The first and second steps deal with determination of relevant management
zone size and the third step concerns determination of the robustness of the
decision. Each step is described in
detail.
Step 1:
Initial ideas for management zone delineation include the use of uniform
grids of various sizes, soil type designation and soil property determination.
The use of electrical conductivity in the definition of appropriate zones
of management, in conjunction with research by Mueller, provides some promising
potential. Delineation based on
soil type, topsoil depth and other characteristics also merits consideration.
Further identification of strategies for management zone delineation will
be completed through additional communications with experts in precision
agriculture including research faculty, extension faculty, agribusiness industry
representatives and producers. More
extensive literature review will also be undertaken.
Step 2:
A mathematical programming model embodying the economic decision
framework of a representative Kentucky crop producer will be formulated.
The objective function will be to maximize net farm returns above
selected relevant costs. Decision variables will include application rates of relevant
nutrients by management zone as well as selection of the number, position and
size of management zones using soil test information from each management zone.
The model will allow for the proper identification of the profit
maximizing management zone for each individual input applied (as determined by
the available database). The
best management zone for one input will not restrict derivation of the best
management zone for another input. Constraints
modeled will include environmental impacts, land available, input purchases,
commodity sales and management zone calculations.
Data required include yield results by cell,
soil test grid samples, fertilizer application rates by cell, commodity price,
fertilizer prices, soil sample test cost and area per cell. Appropriate procedures (such as biophysical simulation or
statistical regression) will be used to develop production response functions.
Biophysical simulation can be used to estimate the underlying crop yield
for a constant technology by altering production and management practices.
Alternatively, underlying production functions could be estimated
assuming various forms from an actual data set if available from an actual
producer that possesses a limited number of but sufficient quantity of
observations. Yield could be
estimated as a function of the level of nutrient to be variably applied,
selected soil properties, selected weather variables and the like. As required,
appropriate testing (multicollinearity, autocorrelation, heteroskedasticity and
examination of alternative production function forms) would be undertaken.
Supporting data (yields, soil samples, etc.) from producers and colleagues
will be requested as a preferred choice with biophysical simulation serving as a
second alternative if such data is unavailable.
The most appropriate results will be
incorporated into the economic model to allow for projection of yield results
for each management zone as based upon average fertilizer application. The model will therefore determine the management zone that
each individual cell should optimally be allocated within. Given the cost of a soil test for each zone, additional costs
for increasing the number of zones is weighed against the yield result
differentials for added accuracy. Thus,
an assessment of the relative performance of alternative strategies for
delineating management zones can be performed.
Step 3:
The sensitivity of the net returns and the chosen optimal management zone
to changes in the economic environment is investigated through alterations in
the economic model. Systematic
increases and decreases (e.g., 5, 10, 15%) of selected economic data will be
included in the model and the new optimal solution determined.
The economic components altered will include commodity price, fertilizer
prices and cost of soil sampling. The
sensitivity of economic results associated with different attitudes towards
production risk would also be appropriate.
E-V, or mean-variance, analysis is a widely used and accepted method for
analyzing risk which could be used if supporting data is available.
Expected
Benefits
The
purpose of this study is to establish a fundamental framework which will permit
analysis of a very germane and basic question currently plaguing the successful
implementation of precision agricultural variable rate technology: How does a
producer identify the optimal management zone? The innovative model formulation
proposed within this research project permits the appropriate economic analysis
to be conducted for comparison to the less data intensive and farmer friendly
management zone delineation procedures presently being conceived and tested by
others (such as delineation by soil properties or electrical conductivity).
Therefore, this research also provides complementary economic comparison among
these alternative tools for management zone delineation.
Additionally, the techniques include the economic assessment of another
very important question: What is the optimal uniform grid size?
Furthermore, this research will provide insights into the establishment
of practical and simple decision rules that may be used to this end by producers
using precision agriculture.
Ultimately,
this research aims at providing the missing element to permit economic
comparison of alternative procedures for management zone delineation and for the
improvement of these decision rules. Dissemination of the decision rules and
their performance is anticipated through web pages, popular articles, extension
materials, presentations and field days.
Deliverables
The
many deliverables of this research include the establishment of an economic
model that can be used in making economic comparison of proposed management zone
delineation techniques. Actual comparison of alternative decision rules will be
made for case studies. Model
results could provide insights useful in the refinement of current and
developing management zone delineation rules. It is also hoped that model
results lead to the development of actual decision rules if the need arises.
Refereed journal articles, popular articles including materials for web
pages, presentations and extension materials including items suitable for field
days and decision support tools are the physical deliverables anticipated from
the project. These different
outlets are used in order to gain the greatest exposure to different audiences:
other researchers and extension specialists, county agents, producers and others
directly. Additionally, it is noted
that in today’s electronic age it is necessary to provide information via the
Internet and such will be done through various web sites (e.g., University of
Kentucky College of Agriculture). Presentations,
professional meetings, and in-service training with county agents will be
provided to facilitate dissemination of information directly to producers.
Furthermore, communication of results through extension newsletters,
extension publications and field days is envisioned.
While
direct model usage is inappropriate for most producers and the underlying
production functions are farmer and area dependent, the establishment of a
framework of economic analysis represents the first stage of solving the problem
of optimal management zone delineation. Furthermore, a means of evaluating
practical, farmer oriented decision rules will therefore be available as well as
establishing potential rules of thumb.